If you and he are saying that understanding depends on concepts - involves classifying input in terms of concepts) – I agree. (That’s not the way you actually defined it, even if you may have meant that).
But concepts do not involve – and are antithetical to – databases of any kind – and are also antithetical to “facts”. Every concept is free-form/informal and general – the essence of GI. Databases and facts are set form/ formal and specific – the essence of narrow AI. And AFAICT from these discussions, you Aaron and others are wasting your time using any kind of database or semantic net or frame for AGI. And to restrict understanding to – or centrally focus it on – language, as Hausser does, is absurd. Obviously, understanding/conceptualisation is necessary for ALL forms of sensory and information-processing. Obviously, language is from an evolutionary perspective, an extremely belated and NOT necessary or fundamental application of conceptualisation. Animals successfully deal with the world without language – and we will be doing incredibly well, if we can just begin to emulate them. Obviously, the foundation of all real world forms of intelligence is actually dealing with and understanding the “movie” of consciousness – understanding scenes of images, not texts of symbols - and not just conceptualising,but simultaneously, interdependently, simulating/enacting what we see. . A robot that can pass the Woz test and GO TO THE KITCHEN will primarily have to deal with “movie (not textual) problems” of understanding, as it navigates a house and kitchen. And I seriously doubt whether we will instruct it using any form of symbolic language - we will use “sign language” – as we do with animals. From: Piaget Modeler Sent: Thursday, April 04, 2013 5:02 PM To: AGI Subject: RE: [agi] Step One towards the real lingua franca of brain/AGI I'll repeat myself again using different words for Mike Tintner's benefit... Understanding is the construction of concepts forming a mental model (a database of "facts"), such that the model can be activated by sensory stimuli to recognize signs of language , and can be used to generate signs of language. (This is actually Roland Hausser's Database Semantics definition). Cheers, ~PM -------------------------------------------------------------------------------- From: [email protected] To: [email protected] Subject: Re: [agi] Step One towards the real lingua franca of brain/AGI Date: Thu, 4 Apr 2013 16:36:09 +0100 PM, Like your exposition of your work which we discussed a while ago, your statement below doesn’t deal with the problem to be solved – in this case: what is the “language” of language (and AGI)? [It’s a controversial but stimulating assumption that there is such a thing as a common “language” – or, to use another metaphor, “currency”.] I’m suggesting that the use of sign language – the use of hand “graphics”/”figures” - is one clue to that lingua franca. Your definition of “understanding” is essentially a non-definition. It doesn’t explain what understanding *entails* – merely points out one v. narrow *application* of understanding, i.e. to language. Obviously, if you think about it, we also have to “understand” what is going on in a visual scene, or indeed understand sensory images of all kinds, including paintings, cartoons, maps, blueprints, x-rays, music and many, many other things. Understanding applies to not merely registering, but successfully classifying EVERY form of input to a real world agent’s brain, not just one. What I’m proposing is that there may be a common form of “language” to all or nearly all these forms of process - a “language” wh. is actually radically different from the purely symbolic kind to which most AGI-ers cling very unimaginatively (in all senses). From: Piaget Modeler Sent: Thursday, April 04, 2013 4:15 PM To: AGI Subject: RE: [agi] Step One towards the real lingua franca of brain/AGI I feel like I'm repeating myself: #7 - Understanding is learning a new language to the point of fluency. (When the words in the new language activate your language independent concepts, and you have created sufficient behaviors so that you can effortlessly generate expressions in the new language). Even when the new language is a signed language. ~PM -------------------------------------------------------------------------------- From: [email protected] To: [email protected] Subject: [agi] Step One towards the real lingua franca of brain/AGI Date: Thu, 4 Apr 2013 12:16:06 +0100 Language by mouth and by hand April 3rd, 2013 in Other Sciences / Social Sciences Humans favor speech as the primary means of linguistic communication. Spoken languages are so common many think language and speech are one and the same. But the prevalence of sign languages suggests otherwise. Not only can Deaf communities generate language using manual gestures, but their languages share some of their design and neural mechanisms with spoken languages. New research by Northeastern University's Prof. Iris Berent further underscores the flexibility of human language and its robustness across both spoken and signed channels of communication. In a paper published in PLOS ONE, Prof. Berent and her team show that English speakers can learn to rapidly recognize key structures of American Sign Language (ASL), despite no previous familiarity with this language. Like spoken languages, signed languages construct words from meaningless syllables (akin to can-dy in English) and distinguish them from morphemes (meaningful units, similar to the English can-s). The research group examined whether non-signers might be able to discover this structure. In a series of experiments, Prof. Berent and her team (Amanda Dupuis, a graduate student at Northeastern University, and Dr. Diane Brentari of the University of Chicago) asked English speakers to identify syllables in novel ASL signs. Results showed that these non-signing adults quickly learned to identify the number of signed syllables (one vs. two), and they could even distinguish syllables from morphemes. Remarkably, however, people did not act as indiscriminate general-purpose learners. While they could easily learn to discern the structure of ASL signs, they were unable to do so when presented with signs that were equally complex, but violated the structure of ASL (as well as any known human language). The results suggest that participants extended their linguistic knowledge from spoken language to sign language. This finding is significant because it shows that linguistic principles are abstract, and they can apply to both speech and sign. Nonetheless, Dr. Berent explains, language is also constrained, as not all linguistic principles are equally learnable. "Our present results do not establish the origin of these limitations- whether they only result from people's past experience with English, or from more general design properties of the language system. But regardless of source, language transcends speech, as people can extend their linguistic knowledge to a new modality." Provided by Northeastern University "Language by mouth and by hand." April 3rd, 2013. http://phys.org/news/2013-04-language-mouth.html AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
